Advice on Presentations

There is a lot of information on presentations out there – both in general, but even specific to technical domains and CS. This is my attempt to give my current take on it (emphasis on current since its always evolving). I do recommend that you look for other people’s opinions as well (with some caveats and suggestions below).

Key point #1 is that giving a good presentation takes effort. And becoming a good presenter takes effort. As you become a good presenter, it may take less effort to give a good talk, but it still takes effort. Learning about presentation, and keeping your skills sharp, is part of it. Reading about it, practicing (doing it), watching others, etc. is important (suggestions later). Reading my thoughts (and hopefully others) is a good starting place – but the main function of this is to provoke you to think.

I apologize in advance for the disconnected nature of this document. The goal here is to get the ideas down. Then I’ll figure out how to better organize it (ideas welcomed). Feedback in general is welcomed.

One other thing to stress before diving in: this is about presentations overall. How to give a good talk. It’s not about how to make good slides (although, there will be some on that). The important thing is giving a good talk. Good slides can be a part of that, or a tool towards that, but I don’t think its the place to start.

Also, I am the first to admit that I don’t always give great talks. I don’t always practice everything I preach. In fact, my opinions on how to give a good talk are continually evolving. I definitely have my own peculiar style. There are some examples of my talks on the web, so you can see for yourself (lots of my slides at at http://gleicher.sites.cs.wisc.edu/Talks, and there’s even a link to at least one video of a talk).

Your mileage will vary

Key point #2: there is no one right answer. There are many ways to give a good talk in general. For any talk, the “best” way to give a presentation will depend on…

You. Each person is different. What works for one person may or may not work for another person. We all have different styles, abilities, etc. It’s actually good to develop your own personal style. And we have different abilities – if you’re a non-native speaker, your talk will likely be different than someone is.
Your talk goals. This is a point that not everyone appreciates, but how you present should depend on what you are trying to communicate. A big lesson is that knowing your goal is important. Having a goal in mind (“this is what I want the talk to do” or “this is what I want the audience to walk away with”) is key. For a class lecture, I usually would speak of “pedagogical goals” (what I want the students to learn), but for a other talks your goal might be “I want people to think I’m a great researcher / perfect for the job” or “I want people to be motivated to read my paper” or …
Your venue. The talk you are giving matters. A 20 minute conference presentation is different than a job talk is different than a general audience lecture is different than a technical colloquium, etc. Some things are similar, but others are different.
Your audience. This is related to venue. But you should consider who is listening (what are their backgrounds, goals). Non-native listeners have related issues to non-native speakers. Even the amount of audience attention can be factored into your talk: there are some audiences that you know that most people in the audience will be more focused on their laptops or text messaging than you.
Your material / topic. If your material is dense and mathematical, your talk may be different than if its a quick survey.
The field. In different communities, there are different standards for presentation. For example, in the humanities (sorry for picking on this one, but its one I’ve actually had some discussion about with people whose opinions I trust) it is often expected that you “read your paper” (how can you improvise – each word should be carefully selected!). These talks (often) fall flat with other audiences. I don’t know if the converse is true: if you give a dynamic, spontaneous, flashy talk to humanities scholars they might react badly. To be clear, within the different genres (e.g. humanities “readings” version computer science “improvisings”) there is good and bad. I generally prefer the genre of talks in my field, but that could just be what I’m used to.

The reason this point (things vary per person and per talk) is so important to make clear upfront is that whatever I tell you needs to be adapted to you and your situation. This isn’t a set of rules to file. Its a set of guidelines to make work for you.

The Key Points

I felt like I needed to give you the first two key points before listing some more, since they are important for interpreting them. Note that some of the later key points can be corollaries to the earlier ones. These will get explained later.

It takes effort to give a presentation, and to become a good presenter.
Your mileage may vary: each presenter and talk is different. Do what’s right for you and your talk.
Respect your audience.
Your visual materials (slides) are not your talk.
Your talk is a performance. Think of it as performance art.
Be Yourself and Have Fun.
Watch good speakers, but remember #2. You can learn from bad speakers as well, but its not as much fun.
Your voice is an important instrument. Timing matters.
Organize and structure for the audience.
Know your goals.
Use fewer words.
Keep the “onion” model – have reconnect points.
Talk to your audience, not to your slides.
A recording of the presentation is not the presentation.
A job talk is different

OK, there’s probably more, but those are good ones. Notice that there isn’t too much about the details (how should I make my slides? when should I use a laser pointer?), since those details should all fall out from the more general principles.

Other Resources

You’d think this would go at the end, but it’s a good way to introduce some of the issues.

For someone else’s thoughts, check Jonathan Shewchuk’s Giving an Academic Talk http://www.cs.berkeley.edu/~jrs/speaking.html. Many of the thoughts here come from there (like the Onion model). One of the best pieces of advice that Jonathan had isn’t on this page. He said (and this is a paraphrase from an email 4 years ago) “you just have to decide you want to be known as a person that gives good talks, and put the time and effort it takes to do it.” There’s a big lesson in that – it really lead me to Key Point #1.

There are lots of good resources on how to design slides. There are books, blogs, web pages, etc. Its all useful. Even the bad stuff can be useful for thinking “why don’t I like this.” But its not where you should start. Start with deciding what talk you are going to give and how you are going to give it. Then worry about how you will support it.
Figuring out how to make slides (and the graphic design issues) is not the place to start. (more on this later)

One of the best ways to learn is to watch others. (this is key point #7). This isn’t as easy as you might think since truly great speakers make things work at a level that you might not notice – they make it seem effortless. Ideally, you aren’t paying attention to how they are being good speakers, but rather to the content/message. But by really attending to how they make it work, you can learn a lot.

In watching others, remember, your talk will be different. You can watch Steve Jobs, Hans Rosling, Barack Obama, Jerry Seinfield, etc. – and you’ll even find blog posts like “how to give a talk like Steve Jobs” and the like. But remember: you aren’t them, your audience isn’t their audience, etc. That said – you can learn a lot. Listen to a political speech (by a good speaker) and see how good speakers can use their voice (timing, volume, pitch, …) to control their message. Some of these people are masters. Its unlikely you will be as good, but it will give you something to strive for.

I like the Presentation Zen blog as well, although I don’t read it as often as I probably should.

Respect Your Audience

Key Point #3, really may be “the golden rule” – from which all else follows. Your audience is giving you their valuable time and attention. You have a responsibility to provide them with a good experience in return. You owe it to them to give an appropriately good talk.

Part of respecting the audience is to talk to them at the appropriate level. Don’t try to bury them in something they won’t understand, or read to them what they could read for themselves, …

Above all Else?

Be Yourself and Have Fun.#6.

The most contagious thing is your passion. If you are excited about what your talking about, your audience will be too. If you bore yourself, just think what that does to your audience.

On Slides and Materials

Notice that I did not start with telling you how to make your slides. If you start with slides, you’ll probably end up in the wrong place.

Let me start by considering what materials will be used for. Note that there are (at least) 3 very different uses of materials:

The thing the audience sees during the talk as a “prop” (to augment the presentation).
The notes the speaker uses to remind themselves what to say.
The materials the audience takes home to remember what was in the talk.

My conjecture is that these 3 things are quite different, and that it is difficult (if not impossible) to create materials that effectively serve all three. In particular, I strongly encourage you to separate #1 and #2. If you don’t it becomes a disaster in many ways: you look at your slides (and not your audience), you read your slides (that the audience can be reading), you tend to be redundant (the audience can figure out your talk without you), …

One thing that I’ve tried: use powerpoint to create #2 (or #3). Then go back, and make a different set of slides that supports what you think you’ll be saying based on what you made first.

Good slide design is important. But I think the key thing in good slide design is meeting some of the other things we’ll talk about below (talk organization, attention control, …). So we’ll come back to this.

Talk Organization: What is your story?

There’s a lot to say here. But I don’t even know where to begin. That’s instructive, it tells me where to begin: know what your message is, and everything else revolves around that.

People will probably take away 1 idea. Remind them of it at the end. I like to think of it as if someone who saw your talk runs into someone who didn’t in the elevator, they need to have the 1 sentence summary.
If you aren’t clear yourself what the 1 sentence takeaway summary of the talk is, you aren’t ready to be giving it.

Good talk organization is kindof like organizing anything else (like writing), but not quite. Here are some thoughts:

There is more of a pressure (for most venues) to motivate the listener to give you their attention.
Because of the abbreviated nature of the timing, you may need to be more explicit about what the audience is supposed to learn: they don’t have time to think and chew on each piece – you need to tell them why things are good / important / …
The audience can’t go back and refer to other parts of the talk (as they could with a paper).
Some things work well in talks (giving the overall motivation and intuition, showing results), some things work better in the paper (giving technical details, working through equations). Be realistic about what you try to get into your talk.
I think its better to make a smaller number of points well, then more points less well.

Re-Connect Points

This is a special issue in organization, but I feel its so important it must be called out. It really sunk in from reading Shewchuck’s notes (see above).

You should assume that your audience will get lost from time to time. You should have points in your talk where you can re-connect with them. Or at least a chance for those who are following you to catch their breath and realize they still are with you.

What you don’t want is for someone to get lost briefly (because they didn’t understand something, or sneezed, or checked their email, or …) and basically miss out on the whole rest of the talk. People will get lost. You job is to bring them back in, preferably before too long and you’ve lost them for good.

Here are some examples of re-connect points:

Explicit “section beginnings” – where you can (effectively) say “we’re done talking about that last thing, and we’re starting to talk about a new thing.” That way, if someone missed the last thing, they might be able to come back and hear the next thing.
Section summaries – where you briefly summarize what happened in the previous section, to give people who missed it a little bit of it to prepare them for what’s next.
An overall talk outline periodically – show people where you are in the talk, and possibly recap how you got to where you are.

Some of this “signposting” adds inefficiency to your talk. You will say the same thing multiple times. You will summarize or preface. You will give framing (that isn’t necessarily content). But this is really important: people will not hear everything you say and keep up with you 100% of the time. You need to make sure that people get value from your talk, no matter what.

Note that re-connects suggest that certain talk structures are better than others (see the “Onion Model” in Shewchuck’s notes). If you have a talk where each piece builds on the previous one, and missing one step means the listener is lost, then it will be harder to re-connect with listeners (Shewchuck calls this the “clew” model). In contrast, a better organization is one where you give the main message at the beginning, and then give more and more layers of details around that core. Like layers of an onion.

The Audience’s Attention

When you’re giving your talk, there are a few places that the audience’s attention will move between:

You.
Your slides.
Their email. (or whatever they do on their laptops)
Other people.
Daydreaming / thinking about something else.

There are probably some others, but you get the point. The main ones to think about are the first two (you vs. your slides). But, here’s something to think about: if the audience is busy reading your slides, they are less likely to be watching you / listening closely to you. Its difficult to know if you want the audience’s attention on you or the slides. Some of it depends on the style of talk, the venue, and the goals.

TED talks are often really good talks, and they are usually well-produced talk videos. But here’s an interesting thought: the camera either shows the speaker, the slides, or a combination (with an occasional audience inset). On one hand this is “the video of a talk is not the talk” – but it’s instructive. Where did the video producer think your attention should be and how did they know? If you were there, would the speaker have been doing something to help you make the right choice?

The other things that are not your talk (notably the laptops) are out of your control. Be ready for it. I call the people who are focused on their laptops “gophers” since they will have their noses buried in their laptop and periodically stick their head up (like a gopher poking out of its hole). My suggestions to deal with this?

Re-connect points become even more important
Use intonation / volume / pauses to signal to the audience when something important is going to happen.
Make sure to “announce” your slides (more on this in a moment)
Design your slides for skimmability. A gopher can take a quick look, decide if they are interested, and listen if they are (or go back into their hole until the next reconnect point).

Delivering the Talk

I can’t over-emphasize how important the “talk” part of the talk is. The way you speak, the way you use non-verbal cues to direct the audiences attention, the timing, how you use pitch and intonation to set aspects of the talk apart, …

A talk is a performance. Your performance is important.

I don’t know how one learns to do this, or even how to explain it. But here is a mis-mash of thoughts:

Use timing to “announce” slides. Generally, you want to pause when you switch slides. The audience WILL read ahead, and not pay as close attention to you. Also, its a good chance to let them adjust to a new thought. The exception is when you want a slide to occur mid-thought/mid sentence. (changing visuals precisely timed with what you’re saying). This can be powerfully effective, but loses its power if its overused. The element of surprise (especially if you’ve been good about pausing) is powerful.
Use timing, volume, pitch, … to emphasize things.
(added Oct 2011) – Use hand gestures effectively. This is a noted difference between professional speakers and others. The fluency of how gestures connect seems to be important – talk show hosts have a stream of gestures between “rests”. (see Neff’s paper on gesture animation).
Silence is useful. Take a breath between sections. It gives your audience time to take a breath too. Ideally, you breathe at places you want the pauses.
Do not talk to your slides. Face your audience and talk to them. Its OK to refer to your notes, but your slides have heard your talk already.
Making connection with your audience (eye contact is key) is surprisingly important.
Personally, I dislike laser pointers. They will wobble, be distracting, and can generally be done better using good slide design (put the highlighting in the design) and other gestures (unless the screen is huge and far away).
Use fewer and smaller words. Big words and complex sentences are easier to be misunderstood.
If you need to use a word that is likely to be mis-heard (since it is unfamiliar to the audience, because you have trouble pronouncing it, …) put it on a slide to emphasize it. And emphasize it verbally (at least the first time).

And I agree with Shewchuck: thank your audience for their attention. (although, you don’t need to write that on a slide)

Designing Slides

The specific issues in graphic design of slides are dealt with elsewhere. Everyone has an opinion. If you look at the history of my slides, you can see that my style changes over time.

Generally, the bigger picture issues drown out the graphic design issues. If you are using your slides as visual support for what you are going to say, you are less likely to create dense text that your audience will be reading along with you.

Bullet lists get a bad rap. Some of it is well-deserved, since often a bullet list is really an outline to remind the speaker what to say. But overall, bullet lists are the easy way out: a quick way to make something on a slide (as opposed to really thinking about what you want to be saying and designing proper visual support for it). Rarely are bullet lists the best design solution: but they are easy. If you look at my talks and see bullet lists, its usually since I am too lazy to do something better. (often, I’ll start with bullet lists and convert to other things as time allows)

Bullet lists aren’t necessarily evil. They just are usually used badly, and are almost never the best design solution.

In general, pictures are good. They do things that words can’t (if you could say it in words, you could be saying it). But even pictures can be overused. Some pictures thrown in to liven things up are OK. But if you have too many, it can dull the imagery that really communicates your message.

Some redundancy between what you’re saying and what the audience is reading is good. Let them read ahead to see if they want to listen closely. Let them learn the big words before you have to pronounce them. If you’re a non-native speaker (or the audience is non-native listeners), the extra help of redundancy between what is spoken and written is valuable.

Economize: for everything on a slide, ask yourself “what is this doing here” and “what is the audience getting from this.” This doesn’t always lead you to minimalism.

The Graphic Design

I am not a good graphic designer. And my feeling on the aesthetics of slides is always changing (right now, I am into high-contrast – usually black backgrounds, or full-bleed images).

Learn something about graphic design. It will help you. Pick up ideas from looking at good slides from others.

But some “rules” that are probably so obvious you’ll be amazed how often they get broken.

Use big enough fonts/images. Look at your slides from far away (the back of the room). We used to have the rule of putting an overhead on the floor – it should still be readable. Now, look at your laptop from far away. And don’t assume everyone has perfect eyesight.
Don’t overuse variety. Some variation is good, but being consistent with fonts/colors helps someone follow. If everything is emphasized, nothing is.
Be careful about contrast. Text should be crisp and readable. Text over a picture is a nice effect, but can be tricky to get right.
Space is valuable. Use it well. The empty space on a slide does a lot to lead the eye.
Use good fonts. I wish I could tell you what these are.

I could keep going, but you are probably better off learning this from someone more qualified (and possibly through experimentation).

But a pet peeve: you don’t need the conference logo on every slide. People won’t forget where they in the middle of the talk. Having your “branding” on the slide (like a logo) is a little more forgivable, since you may want to pound it into people’s heads.

The Academic Job Talk

Academic job talks are an unusual venue. Even if you aren’t going to give an academic job talk, understanding them is useful since you might need to listen/evaluate one, and that some of the issues (and their solutions) may be generally valuable. A nice article about academic job talks (more general than CS) is here – I think it basically echos what I’m saying.

For many reasons. There are a different set of rules that apply, for a number of reasons:

Your goal is to sell yourself and say how great you are. But you have to do this by talking about something you did.
The audience is diverse, yet technical. You need to impress both people who know your area, as well as people outside it.
You need to show off how smart you are and have some things that are hard.
You need to show off that you are a good speaker, and can be trusted in front of a classroom.
You need to talk about your vision. And how what you’ve done can be a good predictor of your future success.
Too much weight is often placed on the talks.

So, some tidbits:

Have something in your talk for everyone. This implies that for everyone, there will be something that is not for them. You need to have some basic overview (for the people outside your field), and some hardcore technical piece for the person in your area. This makes re-connect points really important.
Have at least one place where you go deep and show something where you do something hard. You not only need to show something hard, but need to say why its hard.
Be sure to motivate your problem, and explain why your solution solves it.
Explain why what you did was hard. Why did solving your problem require a genius.
When you show off results, be sure to tell people what they should be looking for since they may not know.
The one sentence to remember thing is big. You need to have the thing your remembered for in the discussions.
Its good to be clear what you did. If you were part of a team, say so, but give an idea of what your role was.